that seems reasonable to me. To recontextualise it in software terms, the requirement comes from the state but the solution comes from the local level. As long

that seems reasonable to me. To recontextualise it in software terms, the requirement comes from the state but the solution comes from the local level. As long as the requirement isn't overly prescriptive (e.g. "people who can't use their legs are entitled to a wheelchair" vs "local authorities must invest 50M SKR in departments X, Y and Z") that's a good solution. My aversion to the idea of a central state (though I don't have a proposed solution to this) is that because there is no antagonist to push back on poorly thought out requirements, states trend towards centralised prescriptive solutionising which I believe to be the root cause of the problem with centralisation. A classic example was the Soviets as you mentioned earlier, with their economic planning committee (Gosplan) explicitly telling factories how much steel to produce and who to send it to. It's one of those problems that pops up in different areas like software development and management practices too. Markets are good at preventing this because they separate the requirements (the customers) from the solution (the producers) and provide a Darwinian process to optimise for the best performers, but markets have their own problems and inefficiencies that make them a poor fit (like markets' tendency to ignore customers with unique or rare requirements, meaning people with rare diseases or multiple problems get rejected or priced out by medical markets). There are market-oriented solutions that libertarians propose, such as evening out costs through medical insurance, but markets will always leave cracks people can fall though (such as those who cannot afford insurance, or immigrants with pre-existing health conditions) and introducing state mandates into the market often fundamentally break the market itself (see: US health insurance market). State legislation leaves cracks too when written badly, so neither is a perfect solution in practice. I don't have a solution to these problems because I think they're rooted in patterns of thinking rather than systemic design, but they're interesting to think about.